African-Americans have an increased burden of both diabetes and diabetic complications. Since many African-American patients have poor metabolic control, novel interventions are needed, especially for urban patients with limited resources. Recent critical findings in approximately 5,900 new patients in the Grady Diabetes Unit: (a) At presentation, HbA1c averages 9.2 percent -- similar to DCCT "standard therapy" patients. (b) 1/3 of patients already have renal damage, and 1/4 nave retinopathy. (c) A stepped care strategy improves metabolic control: HbA1c falls to approximately 7.5 percent after 12 months. (d) The Diabetes Unit can support good primary care: after one month in the Diabetes Unit, subsequent management in the Medical Clinic lowered HbA1c from 9.0 percent to 7.7 percent in 12 months. However, most of the patients at Grady do not receive specialized diabetes care-typical of patients across the U.S., HbA1c in our primary care sites averages 9 percent. We believe that (a) Diabetes Unit approaches can be translated into program interventions which will be effective in the primary care setting, and that (b) success will require a novel partnership between specialists and generalists -- with specialists contributing to improvement of care for patients they do not see. We hypothesize that provider support strategies -- assisted by the Diabetes Unit -- will result in effective diabetes management in primary care sites. Strategies will be piloted in the Diabetes Unit, and hypotheses will be tested in the Medical Clinic. In a randomized, controlled trial, clinic units will receive (1) usual care, (2) computerized flowsheets, with reminders to indicate a need for evaluation and therapy, and/or (3) directed discussion by endocrinologists, reviewing management and explaining approaches to improve care. We will measure outcomes related to microvascular disease (HbA1c) and macrovascular disease (BP, lipids), assess proteinuria as a specific complication, evaluate potential alterations in provider outlook and patient participation in decision-making, and explore impact on quality of life as well. Thus, we will compare two readily generalizable program interventions - - one based solely on computerized technology, and another involving directed discussions with endocrinologists (offering opportunity to impact provider attitude as well as adjustment of therapy). Our interventions should delineate approaches effective in a primary care setting, as needed to improve care and prevent complications in urban African-Americans with NIDDM.